1. This problem can cause the backup process to mysteriously fail without throwing any errors, as backupninja's "is it time to run the backups?" routine always returns false if there's a file named (for instance) "a" in /root. This bit my organization, causing us to have a significant gap in our backups; if we'd had a disaster while this was going on, a great deal of work would have been lost.
2. The bug is fixed by properly quoting three [:class:] instances; this has been out in Gutsy (and now Hardy) for nine months now; it was fixed in 0.9.4-6ubuntu2, which is one release after the version in Feisty (0.9.4-6ubuntu1).
3. A minimal patch to the one affected file is attached.
4. The bug can be easily reproduced on Feisty by running "touch /root/a" and noting that no backups ever run without being forced to (using backupninja -n).
5. The regression potential is minimal: there have been no side effects reported from the patch's application to the current version in the last nine months, and the change will, in the vast majority of cases, function identically. it has also been applied upstream for at least six months ( http://code.autistici.org/trac/backupninja/ticket/11 ). It only makes the package function consistently in the case that a specially-named file happens to be present in /root.
From https:/ /wiki.ubuntu. com/StableRelea seUpdates# head-a7a957d3d6 91c2754ba24e508 5481107ed703b49 :
1. This problem can cause the backup process to mysteriously fail without throwing any errors, as backupninja's "is it time to run the backups?" routine always returns false if there's a file named (for instance) "a" in /root. This bit my organization, causing us to have a significant gap in our backups; if we'd had a disaster while this was going on, a great deal of work would have been lost.
2. The bug is fixed by properly quoting three [:class:] instances; this has been out in Gutsy (and now Hardy) for nine months now; it was fixed in 0.9.4-6ubuntu2, which is one release after the version in Feisty (0.9.4-6ubuntu1).
3. A minimal patch to the one affected file is attached.
4. The bug can be easily reproduced on Feisty by running "touch /root/a" and noting that no backups ever run without being forced to (using backupninja -n).
5. The regression potential is minimal: there have been no side effects reported from the patch's application to the current version in the last nine months, and the change will, in the vast majority of cases, function identically. it has also been applied upstream for at least six months ( http:// code.autistici. org/trac/ backupninja/ ticket/ 11 ). It only makes the package function consistently in the case that a specially-named file happens to be present in /root.